I used to have this 65c816 fantasy. That Atari or Commodore had put out a machine in this lineage rather than jumping to the 68000. On paper it looks appealing -- 6502 compatibility, 24-bit address bus, super fast interrupt responsiveness, very cycle efficient.

So I wasted some time last year futzing around on the C256 Phoenix project -- wrote an emulator for it (or some aspect of what it looked like on paper) since I didn't have a board yet, wrote some code for it, had a falling out with the creator, a bit of a saga. In any case, what I learned in this process is that I really don't like the '816. It was a hack bolted onto the side of the 6502. Bill Mensch couldn't or wouldn't add new addressing modes and register sets to the instruction set to allow 8-bit and 16-bit to co-exist, so there's a constant awkward mode switch back and forth in terms of how the registers are viewed. Even worse, it has a 24-bit address bus but no registers can deal with 24-bit pointer values, so a constant awkward play with using indirect addressing or using multiple instructions to modify addresses or switch pages and so on. And then on top of that the physical chip doesn't break out all 24 address lines, so interfacing the thing is a pain.

The 816 is still produced 35 years later, and available in PLCC and SMT form factors. But they've had this 35 years to make a version of the chip that breaks out all the address lines instead of multiplexing it, which would make sense with the higher pin count available in SMT. But they haven't done it. Why? They went through the effort of making a microcontroller version of it that actually does break out all 24 lines, which nobody really uses. But no MPU.

In the end the 816 was a bit of a hack done especially for Apple. I understand why Mensch did it, but they could have done better.

In contrast, the Z80 has continued to evolve. You can get a brand new eZ80 etc. with large address space and higher clock speeds and excellent cycle efficiency, etc. and not have to muck around with demultiplexing the address bus, etc.

So all said I now understand why engineers at Atari and Commodore and Apple said "f it" about the 6502 and switched to the 68000. The real tragedy of the late 80s/90s era was that Motorola pissed away the excellency of 680x0 by jumping first to the 88000 (failure) and then the PowerPC (less of a failure but not a success for Motorola). Meanwhile Intel kept backwards compatibility and squeezed every ounce it could out of the existing x86, and that's one reason they did better and Wintel maintained dominance.

After spending time with the eZ80, I have to say most of your complaints with the 816's addressing modes are also concerns on the eZ80. The only way to address 24-bits of RAM is in ADL mode, which requires a specific mode switch, and then that screws up interrupts if you stay there. Further, code density is much worse in ADL mode, so staying there most of the time will actually make 24-bits of addressable RAM a problem. There's no provision for doing I/O block transfers in the ISA, so that makes data transfers on that half of the bus... questionable as to why that bus is even present.

Aw, that's sad to hear.

In retrospect Motorola's 6809 (and then the 6309) probably did the best job of adding 16-bit (and even 32-bit) support to their register set. If they had taken it further and added a 24 or 32 bit address bus, that would have been nifty.

I believe the Acorn Atom nearly used the 6809 CPU. (Vague recollections from a very long time ago).

65c816 was as bad as 8088, didnt stop IBM from steamrolling whole PC industry with it. Commodore switching to 68000 threw away their existing 30% market share and wast software library out the window. At the very least they should of embedded compatibility mode into Amiga computers to let users migrate from C64 overnight.

A Z800 type design that came out in 1980 probably would have sucked too. A 16 bit 6502 shouldn't have sucked because the 6502 is a nice simple micro, but Z80 is a lot bigger, and making a 16 bit version of it in 1980 with the transistor and packaging budget of 1980 probably would have had a whole bunch of nasty compromises.

But our opinion as engineers doesn't really matter much. x86 beat the 68K even though the 68K is a nice chip and the 8086 ... isn't.

In the 80s it beat the 68k by marketshare but didn't completely destroy it. Between the Mac, the Atari ST, the Amiga, and a bunch of Unix workstation vendors the 680x0 had a fairly large market share. But Motorola killed it when they could have evolved it. Intel evolved the 486 to the Pentium, and squeezed everything it could out of the ISA -- but Motorola sidetracked onto 88000 and PowerPC and didn't take the 68k past the 68060. And they dragged Apple with them onto the PowerPC. And honestly, Macs from that era _sucked_ -- PowerPC could emulate the 68k but they crashed constantly and it wasn't just because of the OS's lack of memory protection and the like. The systems were full of bugs because of the architecture switch. Waste of $$ and engineering time.

Also, if you squint right... a "z800" type design was basically the 286/386/486, no? The Z80 was a derivation of the 8080 architecture. 8086/286/386/486/Pentium was, too.

That amount of squinting would wreck your eyesight. The x86 line wasn't even close to binary compatible. The assembly language wasn't too hard to convert, but it still required work.

You mean, like a CMD SuperCPU?

https://www.c64-wiki.com/wiki/SuperCPU

I hear they run Geos great.

If you want a cleaner ROM and a better BASIC, there's the Acorn Communicator: http://chrisacorns.computinghistory.org.uk/Computers/Communi...

Not a games machine, though.

The 65816 was designed to Apple specifications which explains the annoying 64K blocks etc.

IMHO multiplexed address isn't a problem as it's simple logic to fix but having to keep track of 64K segments is more painful

That would be surprising considering Apple didnt know about this processor and had to be convinced to even give 2GS a chance, according to computer history museum Oral History of William David "Bill" Mensch. Not to mention wikis "Steve Wozniak said in January 1985 that Apple was investigating the 65816".

IBM and clones crossed 50% market share in 1986. Commodore with just one model alone, C64, still held ~30% of the market in 1986! http://www.retrocomputing.net/info/siti/total_share.html

All Commodore had to do to maintain the dominance would be respeccing 1985 release of Amiga 1000 to 65c816 and embedding hardware compatibility with C64 (vic/sid on a chip, let c64 programs run embedded in workbench window). Clock for clock 65c816 (8 bit databus!) is almost as fast as 68000(16 bit databus), not to mention stomps all over both 8088 and 8086. http://westerndesigncenter.com/wdc/AN-001_ Instruction_Level_Performance_Comparisons.cfm

Wouldnt be that hard considering Amiga architecture is a big graphics unit with slave CPU glued to its ass.

65816 addressing high resolution graphics that the Amiga had would have been a nightmare to deal with. There are no real 24-bit pointers on the 816, you deal with 64k chunks at a time.

The planar graphics also would have been a PITA as the 816 has no bit manipulation instructions.

Floating point also an issue.

Amiga chipset on 816 with a C64 compatibility mode would have made a nice fast game and home machine but not a real competition for the PC in any kind of productivity aspect.

But I agree the Amiga should have had a C64 compatibility mode. They should at least have offered an expansion card with a C64 on it, like Apple did with the Apple II on a Macintosh. But at this time Tramiel had left Commodore along with Shiraz Shivji and other C64 engineers. So we had this weird situation where the Atari 16-bits looked more like a C64 than an Atari 8-bit and the Amiga looked more like an Atari 8-bit (same chipset designer) than a C64.

I missed the last bit. History is so full of surprise like the future.

By 1986 IBM had a lot more than 50% of the "real" market share and it had all the business software. The C64 was seen as a toy, and while backwards compatibility with it would get you a lot of nice games, that wouldn't help much.

Lotus 123 was released in 1983 and was already dominating by the end of the year.

The C64 wasn't released until late 1982, a year after the IBM PC, and visicalc on a 64K machine can't compete with 123 on a 512K machine.

To "beat" the IBM PC, I'm talking about a high-mem PET-compatible computer in a Vic-20/C64 type time frame. Imagine a Commodore that has 3 models: a vic-20 comparable, a c64 comparable and a $3000 model that supports up to 1Meg of RAM. All 3 models backwards-compatible with the PET and each other. In the 1980-1982 time frame.

Commodore won the low end in the early 80's, but gave up the high end. And Moore's law ensured that the high end destroyed the low end over time.

> And Moore's law ensured that the high end destroyed the low end over time.

Well, I'd say more the middle road won here; the high end was not PCs, but (UNIX) workstations. IBM PC was after all contemporary to SGI IRIX, SUN-1 (and -2), and Apollo computers.

Actually Commodore ("Commodore Business Machines") had a whole portfolio of "business" machines that followed on after the PET that were a failure -- look at the "CBM-II" machines, which could be had with 256KB of memory etc.. They even built but never sold a machine around the Z8000 that could run Coherent Unix (Commodore 900). All were failures because among other things the people running things there could not understand the concept of backwards compatibility and having a real consistent operating system. They were all basically bespoke machines.

Shiraz Shivji from both the C64 and the Commodore 900 team ended up leaving with Tramiel when he bought the corpse of Atari. He ended up being the architect of the Atari ST.

Creating bespoke machines was the norm. S100 was the only standard with a history. Outside of that, manufacturers didn't do backwards compatibility.

IBM clones transformed the market because the PC standard (with MS-DOS) became a commodity, which meant you could swap software and data between machines from different manufacturers and Compaq made sure could buy the hardware via a network of dealers, like a car - something Apple originally innovated, and the S100 scene experimented with, but never quite mastered.

The compatibility was a Really Big Deal at the time. It was far more important to users than the very obvious technical limitations of the hardware. Not only could you share data, but you could take your disks to a different office with different hardware from a different clone maker and still get work done.

Incidentally, Unix would never have happened in this market because the shell syntax (all of them...) is far too hostile for most users.

"1986 after IBM had already won"

I like your post... Just a thing: the PC architecture won, but IBM lost all control of it.

They tried to take over in 1987 with the PS/2 line and its MCA bus, but the clone army was strong enough to fight, and IBM progressively got out of the PC market... until the sales of their hardware divisions: printer division (Lexmark, 1991), desktop and laptop division (Lenovo, 2005). I don't remember for the PC-based servers, but certainly gone also.

I like to think of it like this -- before the PC we had CP/M and 8080/Z80 machines, and it was a fairly open if not fully standardized ecosystem. IBM muscled into that market and displaced it because they were close enough to those systems to not be too alien, but also sufficiently better to get people to switch and to attract new users. Plus their name and marketing advantages.

But in the end the market and community won out and pushed back into something fairly close to the original CP/M ecosystem, and we got the world of clones.

Another one:

- all PC clones used the (not otherwise impressive) FAT file system, there were only a handful (SD/DD, single, double side, all soft-sector'd and in the early eighties all 5.25") floppy disk formats, while in the CP/M world every vendor used its own from 8" to 3" (yes, 3, not 3.5"). So software could much more easily be, er, "shared".

> Is this a 16 bit or an 8 bit machine?

The linked page answers that. It has a 14MHz Z80 and a separate 16b coprocessor that does as yet unspecified tasks.

I'm (a) surprised it is only 14 MHz seeing that it is running on an FPGA. Secondly, I'm surprised they didn't go with a Z800 or Z280, which run legacy Z80 code and have 16b extensions and an MMU for a larger memory space.

It is hard to follow how to start. Do you know? Seems reach the FPGA board but do you buy it say from digi-key then find ... I am lost.

Need those 1970s FYI things. There is newbie. We are all newbie.

But the idea of openness is great. Just not sure this particular one. Still they have a lot like VM, various social channel.

Just no door in.

To be fair, the machines that took Z80 home micros into the 16-bit era were MSX Turbo-R.

https://www.msx.org/wiki/MSX_Turbo_R

Well, Apple ][ had Visicalc. And they were first with a spreadsheet.

I "upgraded" from an Apple ][ to an IBM-PC in 1983. It was because of the IBM name but also the "build quality". A real "business" keyboard, and a better constructed case, monitor, etc.